sign in sign up
<<
Home , Inbreeding, Inbreeding depression

Heredity 76 (1996) 83—91 Received 9 June 1995

Variation in depression among families and of Clarkia tembloriensis (Onagraceae)

TIMOTHY P. HOLTSFORD Division of Biological Sciences, University of Missouri, Columbia, MO 65203, U.S.A.

Therelationship between the self-fertilization rate of a and the severity of inbreed- ing depression is difficult to predict because the underlying may be complex. However, it is important to understand this relationship, and the degree to which varies among families within populations, if we are to understand the of plant systems. Inbreeding depression was studied in glasshouse trials using seed derived from two populations of Clarkia tembloriensis, Cantua Creek and Idria. These had very different rates of self-fertilization (s) and inbreeding coefficients (F) (s =0.74,F =0.77in the CC-i population; s =0.i6,F 0.10 in the I-i population). between these populations was equivalent to outcrossing within them; there was no evidence for or increased from interpopulation crosses. The more self-fertilizing CC-i population had fewer recessive lethal than the outbreeding I-i population. However, cumulative inbreeding depression at the end of flowering was not significantly different between these populations. Inbreeding has seemingly purged the lethal genes from the CC-i population but overall mutational load, especially in characters that are manifest late in the life history, is still substantial despite a history of inbreeding. Variation in inbreeding depression among families within these two populations was surprisingly large. The range of variation among families was five to seven times larger than the difference in inbreeding depression between populations. Variation in inbreeding depression among families should make these populations more susceptible to the invasion of genes which increase the rate of self-fertilization.

Keywords:Clarkia,inbreeding depression, mating system, self-fertilization.

increased homozygosity can be detrimental if the Introduction at individual loci affecting have a Individualswhose parents are related by descent are synergistic overdominant interaction so that hetero- often less vigorous than individuals whose parents zygotes are always more fit than homozygotes. If are unrelated. The decrease in fitness of inbred inbreeding depression is caused by the expression of progeny relative to outbred progeny is termed deleterious recessive alleles then inbreeding depression. Inbreeding depression is an can decrease the severity of inbreeding depression important phenomenon because it can affect the over successive generations of inbreeding. However, evolution of plant mating systems, the speed and the rate at which deleterious recessives are elimi- efficacy of programmes and the nated from populations and the equilibrium fre- choice of strategy employed to sample or preserve quencies of these alleles depend on several of rare or endangered populations. parameters. The number of loci, the degree The genetic basis of inbreeding depression will and kinds of interactions between loci, the rate of determine how inbreeding depression can coevolve , the strength of selection, and the domi- with the mating system. Increased homozygosity can nance coefficients of genes affecting fitness may all lead to inbreeding depression by two different be important in determining the severity of inbreed- genetic mechanisms. Homozygosity can cause dele- ing depression (Charlesworth & Charlesworth, 1990; terious recessive or partially recessive alleles to show Charlesworth et al., i990, 199i). If inbreeding their effects on the . Alternatively, depression results from the loss of overdominant

1996 The Genetical Society of Great Britain. 83 84 T. P. HOLTSFORD

allelic interactions as homozygosity increases then families for fitness-determining genes may allow the evolution of inbreeding depression will follow a genes for self-fertilization to invade an outbreeding different course (Charlesworth & Charlesworth, population despite high average population inbreed- 1990). If the overdominance is asymmetrical (i.e. ing depression, because genes will become one confers a higher average fitness than its associated with high viability genes (Uyenoyama & alternative) then inbreeding depression should Wailer, 1991a). If there is overdominance for fitness decline to zero as the mating system approaches then variation in inbreeding depression among complete self-fertilization because if homozygosity is families will allow outcrossing genes to become inescapable then the allele that performs better associated with high viability genes (Uyenoyama & when homozygous will become fixed. Symmetrical WaIler, 1991b). The associations between genes overdominance, though unlikely, would lead to an influencing the mating system and genes influencing increase in inbreeding depression as self-fertilization fitness may have as much or more effect on the increases, heterozygotes become rare and both course of mating system evolution as the magnitude homozygotes are equally unfit (Charlesworth & of inbreeding depression (Holsinger, 1988; Ueyno- Charlesworth, 1990). yama & Wailer, 1991a,b,c). The degree to which a history of inbreeding modi- This paper was motivated by these shortcomings fies the severity of inbreeding depression is an issue in our empirical knowledge: does a history of on which empirical data are few and confusing inbreeding result in the purging of mutational load, (Toppings, 1989; Holtsford & Ellstrand, 1990; and is there significant variation among families in Rathcke & Real, 1993; Eckert & Barrett, 1994; the severity of inbreeding depression? I also ask Latta & Ritland, 1994; McCall et at., 1994). No clear whether outcrossing between populations results in picture has emerged on the degree to which muta- more heterosis (the complement of inbreeding tional load may be purged by the increased effi- depression), than outcrossing within populations, or ciency of selection acting on more homozygous whether crossing between populations results in inbred populations. Although selection should be outbreeding depression (Price & Waser, 1979; more efficient at decreasing the frequency of delete- Waser & Price, 1991; Waser, 1992). The Cantua rious recessives when those genes are homozygous, Creek-i (CC-i) and Idria-i (I-i) populations of highly inbred populations may still show substantial Clarkia tembloriensis were chosen for this study inbreeding depression (Karron, 1989; Holtsford & because their mating systems and inbreeding coeffi- Ellstrand, 1990; Johnston, 1992). Purging of delete- cients are quite different (selfing rates, s =0.74and rious recessive genes should be more difficult when 0.16; inbreeding coefficients, F =0.77and 0.10, population sizes are small enough so that drift may respectively; Holtsford & Elistrand, 1989), and increase the frequency of these genes, when the because the genetic basis of the floral traits under- deleterious alleles are more recessive, and have lying the mating system differences in these popula- smaller individual effects on fitness (Charlesworth & tions have been studied (Holtsford & Ellstrand, Charlesworth, 1987; Charlesworth et at., 1990). 1992). Furthermore, even recessive lethals may not be purged unless inbreeding exceeds a threshold value, which may be surprisingly high, even for moderate Materialsand methods mutation rates (Lande et a!., 1994). High mutation Thestudy populations occur 36 km apart in the rates to deleterious recessive alleles would also inner coast ranges of central California, U.S.A. impede the purging of mutational load (Johnston & (Idria-1: 36°33'OO"N, 120°50'OO"W, 457 m altitude, Schoen, 1995 and references therein). Cantua Cr-i: 36°24'30"N, i20°28'30"W, 366 m). Another unresolved issue is how much inbreeding Voucher specimens are deposited at the herbaria of depression varies among families within populations. the Universities of California, Riverside and Inter-family variation in mutational load should be Missouri. The mating system and structure of these critically important to mating system evolution populations is described elsewhere (Holtsford & (Uyenoyama & Wailer, 1991a,b,c). In partially Elistrand, 1989). Ten seeds from each of iS mater- inbreeding populations identity disequilibria will nal parents from the CC-i and I-i populations were develop among loci influencing the mating system sown on vermiculite and placed in a growth chamber and loci influencing fitness (Weir & Cockerham, (12 h light: 12 h dark, 24°C day, 7°C night). When 1973; Holsinger, 1988; Uyenoyama & Wailer, the first set of true leaves appeared, one seedling 1991a,b,c). If inbreeding depression is caused by from each field-collected maternal plant was trans- deleterious recessive genes then variation among planted into a 5 cm square pot and the seedlings

The Genetical Society of Great Britain, , 76, 83—91. VARIATION IN INBREEDING DEPRESSION IN CLARK/A 85 were moved to a glasshouse. The seedlings were root axis. The number of flowers, rather than the transplanted to 10 cm diameter pots when they were number of fruits or seeds, was used as an estimate of 15 cm tall and into 4 L pots when they were reproductive effort because the CC-i population is 30—40 cm tall. highly autogamous whereas I-i sets very few seeds in All plants received three pollination treatments the absence of pollinators. Two lifetime fitness esti- each on at least two flowers: (1) self-pollination, (2) mates were constructed by multiplying the propor- outcrossed pollination with pollen from plants from tion of surviving individuals in each treatment! the same population as the maternal plant, and (3) population by the flower production and by the dry outcrossed pollination with pollen from the other weight of that group. ANOVA, Wilkes's test for population (CC-i x I-i and reciprocal). Flowers to normality, and Tukey's Studentized range tests were be crossed were emasculated before anther dehis- performed using ss (SAS Institute, 1985).Correla- cence. Seif-pollinations were achieved by wiping tion analyses and contingency table tests were dehiscent anthers directly upon receptive stigmas of performed using SYSTAT (version 5.2 for Macintosh, the same plants. Each day anthers from one newly Wilkinson, 1989). opened flower from each plant were placed in a Inbreeding depression was estimated as separate small vial for each population and allowed (5=1—wSf/w0X, where w and w0 are the fitness esti- to dehisce. The pollen was mixed with a toothpick, mates of progeny resulting from selfed (Wsf)and and outcrossed pollinations were made with these outcrossed pollinations (w0). Because there were no mixtures of pollen from 15 plants. Because these significant differences in performance between indi- pollen mixtures contained pollen from every plant in viduals grown from outcrossing within populations the population there may have been a small amount vs. outcrossing between populations (see below), I of self-fertilization in the outcrossed treatments. used the outcrossed within-population performance This would bias the inbreeding depression in a to estimate lifetime w0,. Population inbreeding conservative fashion; the inclusion of selfed progeny depression was estimated by pooling the perform- in the outcrossed group would diminish differences ance of selfed and outcrossed plants from all famil- among the treatments. I believe this bias to be very ies in that population. To test for variation in small because only 1/15th of the pollen in any inbreeding depression between populations and outcrossed pollination was selfed pollen. Further- between families the dependent variables were more, in C. unguiculata, a very closely related log-transformed and analysed by ANOVA. Because which is the presumed progenitor of C. ANOVA detects additive effects a significant cross tembloriensis (Vasek, 1964), selfed pollen sired only type by population interaction on untransformed an average of 15 per cent of the offspring when data would indicate that the difference among cross applied in a 50:50 mixture with outcrossed pollen types (self vs. outcross) varies between populations. (Bowman, 1987). If the same discrimination against A significant cross type by family or population selfed pollen were applicable in my pollinations then interaction using log-transformed data indicates 2 per cent of the outcrossed progeny would in fact significant variation in the ratio of selfed to be the products of self-fertilization (1/15 x 0.15/0.5 = outcrossed performance and therefore indicates whether inbreeding depression varies between famil- 0.02). Seeds from 12 (of the 15 original) maternal plants ies or populations (Johnston & Schoen, 1994). were germinated in the growth chamber as detailed Inbreeding depression was also estimated separa- above and grown in the glasshouse. The randomized tely for each maternal parent. In cases where the block experimental design had the following inde- average of a maternal parent's selfed offspring was 2,CC-i greater than the average of her outcrossed offspring pendent variables: source population (n = — and I-i), maternal parent (n =12,nested within I estimated outbreeding depression as 5=—(1w0/ source population), pollination treatment (n =3, so that (5 is symmetrically distributed and self-pollination, outcrossed within population, and bounded on [-i,1] (Agren & Schemske, 1993). crossed between populations), and block (n =6 Confidence intervals for inbreeding depression esti- glasshouse benches). Three replicates were grown mates of maternal plants were estimated as the 2.5th per bench so that each maternal parent was repre- and 97.5th percentiles of the distributions of 1000 sented by 54 progeny, 18 in each pollination treat- bootstrap replicates. Individuals were resampled ment. The dependent variables estimated were randomly with replacement within each treatment , survival to flowering, the number of and within each maternal parent, until the same flowers produced and dry weight. Dry weight esti- number of individuals was chosen as was in the orig- mates include above-ground tissues and the major inal maternal parent/pollination treatment group.

The Genetical Society of Great Britain, Heredity, 76, 83—91. 86 T. P. HOLTSFORD

= 0.53,P = Table 1). Cumulative Results population (x 0.77, survivorship (including germination to flowering) Germinationrates were 95 per cent for the CC-i was 88 per cent for the CC-i population and 74 per population and 82 per cent for the I-i population. cent for the I-i population. The plants resulting Seeds resulting from self-pollination were less likely from self-pollination were less likely to survive in the to germinate than outcrossed seeds in the I-i popu- more outbreeding I-ipopulation (x= 14.35, lation (= 12.32,P =0.002)but not in the CC-i P<0.001), while pollination treatment did not influ- ence survivorship in the more self-fertilizing CC-i population (x= 1.26,P =0.53,Table 1). Table 1 Germination and survivorship frequencies for Pollination treatment had a significant effect on three pollination treatments (self-pollinated [Selfedi, crossed with pollen from the other population [Cross survivorship x dry weight and survivorship x number btw], and outcrossed with pollen from within the maternal of flowers (F2,82663.61, F21146 =13.87,respectively, parents' population [Cross w/in]) in two populations Fig. 1). Plants resulting from crossing within or (Cantua Creek-i and Idria-1) of Clarkia tembloriensis between populations were not significantly different from each other but seeds derived from self-pollina- Cantua Creek-i Idria-1 tion produced fewer flowers and lower dry weight than either outcrossed treatment (Tukey's test, Dead Germinated Total Dead Germinated Total Fig. 1). The two fitness estimates (survivorship x dry weight and survivorship x number of flowers), were Selfed 9 206 215 55 160 215 significantly correlated with each other in both Crossbtw 11 202 213 27 183 210 P<0.001; r = P = Crossw/in12 199 211 34 179 populations (r =0.87, 0.83, <0.005 213 in CC-i and I-i, respectively). Total 32 607 639 116 522 638 Variation in inbreeding depression was examined Dead Flowered Total Dead Flowered Total by dropping the outcrossed between-population treatment and performing ANOVA on log-trans- Selfed 31 184 215 76 139 215 formed fitness estimates (Table 2). The population Cross btw24 189 213 43 167 210 by cross treatment interaction was not significant for Cross w/in24 187 211 48 165 213 either fitness estimator indicating that inbreeding Total 79 560 639 167 471 638 depression did not vary significantly between these two populations. However, inbreeding depression

Cantua Creek-i

-c C) a) Fig. 1 Effect of pollination treatment V on survivorship (surv) >< final dry weight and survivorship x number of flowers in the CC-i and I-i popula- U) tions of Clarkia tembloriensis, which C differ in their natural rates of self- fertilization, s. The treatments are: outcrossed between populations (cross btw), outcrossed within populations Maternal Parent Maternal Parent (cross w/in), and self-pollinated (self). Data are means±2 SE.

The Genetical Society of Great Britain, Heredity, 76, 83—91. VARIATION IN INBREEDING DEPRESSION IN CLARK/A 87

Table 2 Analysis of variance of log(fitness estimates) for Clarkia tembloriensis. (POPN) indicates nested effects within population. 'Mom' is short for maternal parent. Bench interactions and three-way interactions were not significant and so were dropped from the model

Source d.f. Type III SS MS F

Dependent variable: log[(no. of flowers) (survivors hip)] Population 1 0.04 0.04 0.03 Maternal parent (POPN) 22 32.39 1.47 4.52*** Cross type 1 7.86 7.86 24.17*** Bench 5 0.49 0.10 0.30 Population x Cross type 1 0.65 0.65 1.99 Mom(POPN) x Cross type 21 25.16 1.20 3.68*** Dependent variable: log[(dry wt)(survivorship)] Population 1 5.33 5.33 4.16 Maternal parent (POPN) 22 28.23 1.28 445*** Cross type 1 16.12 16.12 55.96*** Bench 5 1.49 0.30 1.03 Population x Cross type 1 0.17 0.17 0.58 Mom(POPN) x Cross type 22 29.97 1.36 473*** ***Denotes P<0.0001; other effects are not significant.

Cantua Creek-i 0.8- 0.6- 0.4- 0.2-.t 0 (1— ' -0.4- ii it •- -0.6- •-0.8- 4 —I— , II II I I 0.8- C) C) 0.4. 0.2. Fig. 2 Variation in inbreeding depres- z I. sion ((5) among maternal families in two populations of Clarkia tern blorien- •--0.4. sis. Error bars denote the 2.5th and -0.6. 97.5th percentiles of 1000-bootstrap -0.8 distributions. The dotted lines denote 1II 2 345Ill 6I 78I I9101112I II I the population-wide inbreeding Maternal Parent Maternal Parent depression. varied significantly among maternal parents, as ering,to final lifetime dry weight and flower produc- shown by the significant interaction terms for both tion, with one exception (Fig. 3). In the I-i fitness estimates (Table 2), and by the wide range of population, inbreeding depression in survivorship inbreeding depression estimates for different mater- alone could account for the inbreeding depression in nal parents (Fig. 2). lifetime flower production (survivorship x number of The magnitude of inbreeding depression generally flowers). increased from germination through survival to flow-

The Genetical Society of Great Britain, Heredity, 76, 83—91. 88 T. P. HOLTSFORD

Cantua Creek-i 0.4 I I oO..,.-1C .,I I 0C 0 I (0 1) 0. 0)0.2]I V I Fig. 3 Inbreeding depression (6) in C) )0.1-1 C twopopulationsof Clarkia temblorien- as1 sis at different life history stages: I I I 0) vI_iI I germination (germ), survival to flower- CI iiI C ing (surv), number of flowers (no. firs), wt germ surv no.tlrs drywt germ surv no.flrs dry and final dry weight (dry wt).

flower significantly less often than their outcrossed Discussion siblings. Crossing between these two populations, which are Inbreeding depression generally increased through separated by 36 krn, resulted in progeny that were the lifehistory in both populations (Fig. 3). equivalent in all performance measures with progeny Although the more self-fertilizing CC-i population from within-population crosses. Progeny from may have had its early acting deleterious recessive between-population crosses might have exhibited genes purged by natural selection, lifetime inbreed- heterosis because they were more heterozygous than ing depression estimates do not differ significantly progeny from within-population crosses (expected between populations —despitethe almost 5-fold heterozygosity at isozyme loci is 0.23 for interpopu- difference in self-fertilization rate and 10-fold differ- lation crosses, vs. 0.14 and 0.07 for random outcros- ence in observed heterozygosity. The disparity ses within I-i and CC-i, respectively; Holtsford & between inbreeding depression in germination and Ellstrand, 1989). However, if the populations had survival vs. inbreeding depression in lifetime produc- diverged sufficiently from each other, either through tivity may be explained by differences in selection local or by drift, then progeny from intensity. Purging of recessive genes with larger between-population crosses might have exhibited effects on fitness proceeds much more rapidly than outbreeding depression (Waser & Price, 1991; Van the purging of moderately deleterious genes (e.g. Treuren et al., 1991; Waser, 1992). Because I found Fig. ia of Charlesworth & Charlesworth, 1987). no differences between within- and between-popula- Although inbreeding should promote the purging tion crossing treatments it seems likely that a 9 or 16 of deleterious recessive genes the relationship per cent increase in heterozygosity was not enough between the selfing rate and inbreeding depression to cause detectable heterosis, nor was the divergence is not simple. The magnitude of inbreeding depres- between these populations great enough that hybrids sion will depend not only on the mating system but showed outbreeding depression. on the strength of selection on, the Higher homozygosity in naturally inbreeding coefficients of, and the mutation rate to deleterious populations should result in lower equilibrium genes, as well as the degree and kind of epistasis frequencies of deleterious recessive alleles compared among fitness-determining loci (Charlesworth & to more outcrossing populations (Crow, 1948; Charlesworth, 1990; Charlesworth et al., 1990, 1991). Wright, 1969; Lande & Schemske, 1985; Charles- Further complicating the relationship between the worth et al., 1990). Average observed heterozygosity mating system and inbreeding depression are iden- (H0bS) is higher in the outbreeding I-i population tity disequilibria among genes determining the than in the more selfing CC-i population so the mating system and genes influencingfitness frequency of deleterious recessive genes should also (Holsinger, 1988; Uyenoyama & Wailer, i99ia,b,c). be greater in I-i (Hob, =0.10in I-i vs. 0.01 in CC-i; The complicated relationship between the mating Holtsford & Ellstrand, 1989). Early acting lethal system and the severity of inbreeding depression is were more common in I-i. The naturally borne out by the variation in empirical results found inbreeding CC-i population did not show significant in the literature. My previous work found that the prereproductive inbreeding depression in this experi- rank of the self-fertilization rate was correlated with ment or in previous investigations (Table 1, Fig. 3 of the rank of the severity of inbreeding depression in this paper; Table 7 of Holtsford & Ellstrand, 1990). three C. tembloriensis populations (Holtsford & However, plants resulting from self-pollination in Elistrand, 1990; Fig. 7) —butonly for three of four the more heterozygous I-i population survived to measures of 6 and only in one of two experiments.

The Genetical Society of Great Britain, Heredity, 76, 83—91. VARIATION IN INBREEDING DEPRESSION IN CLARK/A 89

Barrett & Charlesworth (1991) demonstrated that inbreeding to spread even if the average inbreeding mutational load decreased during five generations of depression for the population exceeds 0.5 (or what- inbreeding in a naturally outbred population, but ever the critical level might be if pollen discounting not in a naturally selfing population of Eichomia is considered; Holsinger, 1988; Uyenoyama & paniculata. Latta & Ritland (1994) used inbreeding Wailer, 1991a). Identity disequilibria among loci coefficients (F) estimated from isozyme data to ask develop under inbreeding (Weir & Cockerham, whether prior inbreeding was a good predictor of 1973), so that associations among alleles at mating inbreeding depression. Although there were negative system loci and viability loci will not be random. associations between F and several measures of (5, Lineages which are homozygous both for genes onlyone (5 estimate was significantly correlated with promoting inbreeding and for partially dominant F and populations with different inbreeding histories 'good' viability genes should prosper because of the sometimes showed very similar levels of inbreeding higher fidelity of transmission to the next generation depression. Other studies report weak or no associa- of the beneficial viability genes in inbreeders relative tion between mating system and level of inbreeding to outbreeders (Uyenoyama & Wailer, 1991a). depression (Toppings, 1989; Rathcke & Real, 1993; The primarily outbreeding I-i population may be Eckert & Barrett, 1994). The relationship between vulnerable to the invasion of genes which would inbreeding history and the level of inbreeding increase self-fertilization, e.g. via from depression may differ even between different CC-i or a similar conspecific seifing population. performance traits measured on the same plants. Although I-i has a higher frequency of lethal genes McCall et al. (1994) found that serial inbreeding of than the more selfing CC-i population, lifetime (5 Impatiens capensis seemed to diminish the muta- estimates are below 0.5, the critical level of inbreed- tional load in seed weight and biomass but that ing depression that prevents the spread of seifing survivorship became progressively worse as inbreed- genes in simple models of mating system evolution ing continued. (e.g. Maynard Smith, 1978 & model la of Lloyd, Inbreeding depression varies more among families 1979). However, I might have underestimated within populations than between populations. inbreeding depression because I estimated (5 in the Inbreeding depression in the outbreeding I-i popu- glasshouse and because I could not quantify lation ranged from —0.4 to 1.0 among families inbreeding effects in all fitness-related traits, e.g. (Fig. 3) while differences in (5 among populations male reproductive function. were of the order of 0.1 (dotted lines in Fig. 3). Pollen discounting may retard the evolution of Variation between families in fitness between selfed selfing in Clarkia. Selfing populations of C. temblor- and outcrossed offspring has been demonstrated in a iensis produce less total pollen and have lower few studies by a significant ANOVA interaction pollen:ovule ratios than outcrossing populations. between maternal parent and pollination treatment Hence, selfers' potential for male reproductive (Schemske, 1983; Sakai et a!., 1989; Agren & success is lower than that of outcrossers (Vasek & Schemske, 1993; McCall et a!., 1994), but some Weng, 1988). The estimated pollen:ovule ratio of I-i investigations did not find any evidence for fitness is 148:1 while that of CC-i is 73:1 (collections T2 variation among families (Schemske, 1983; Johnston, and T6 of Vasek & Weng, 1988). The critical level 1992). It is not clear how many of these maternal of (5 that would prevent the spread of a selfing gene parent x cross type interactions would be significant would be less than 0.5 if invading selfers suffered if the ANOVA5 had been done on log-transformed not only inbreeding depression but also reduced fitness estimates, as advocated by Johnston & ability to transmit the selfing genes via male function Schoen (1994). (Holsinger et al., 1984). Further complicating the By estimating (5 and its underlying distribution for prediction of the fate of selfing genes is that the each maternal family one can more easily see the floral traits which promote selfing (reduced anther- extent of the variation in inbreeding depression. Five stigma separation and shortened protandry) are not to eight maternal families exhibited inbreeding simply inherited. Segregation of F2 and backcross depression that was significantly different from the progeny from hybrids between I-i and CC-i showed population inbreeding depression estimate, depend- that genetic divergence in mating system traits ing on which fitness estimate and which population between these populations is polygenic (Holtsford & one considers. Family 3 from the I-i population had Ellstrand, 1992). a significantly negative (5 by both performance esti- The marked variation among maternal families in mates. Variation in (5 among families is interesting inbreeding depression should facilitate the invasion because such variation may allow genes promoting of selfing genes. Because highly inbred populations

The Genetical Society of Great Britain, Heredity, 76, 83—9 1. 90 T. P. HOLTSFORD still show significant inbreeding depression (see also Genetics, 33, 477—487. Holtsford & Ellstrand, 1990), it is likely that delete- ECKERT, C. G. AND BARRETT,S.C. H. 1994. Inbreeding rious recessive genes (not overdominant interac- depression in partially self-fertilizing Decodon verticilla- tions) are responsible for the inbreeding depression tus (Lythraceae): population-genetic and experimental in these Clarkia populations. As identity disequili- analyses. Evolution, 48, 952—964. bria are established, those families with selfing genes HOLSINGER, K. E. 1988. Inbreeding depression doesn't mat- ter: the genetic basis of mating system evolution. Evolu- and 'good' viability genes should prosper at the tion, 42, 1235—1244. expense of lineages with other combinations of HOLSINGER, K. E., FELDMAN, M. W. AND CHRISTIANSEN, F. B. mating system and viability alleles (Uyenoyama & 1984. The evolution of self-fertilization in plants: a Waller, 1991a). For example, if a gene promoting population genetic model. Am. Nat., 124,446—453. selfing were crossed into a genetic background HOLTSFORD, T. P. AND ELLSTRAND, N. C. 1989. Variation in similar to that of I-i families 3, 7, 8 or 12 then these outcrossing rate and population genetic structure of lineages should do very well because they seemingly Clarkia tembloriensis (Onagraceae). Theor App!. Genet., suffer no inbreeding depression. Furthermore, the 78, 480—488. genes promoting selfing would increase the fidelity HOLTSFORD, T. P. AND ELLSTRAND, N. c. 1990. Inbreeding with which both the 'good' viability genes and the effects in Clarkia tembloriensis (Onagraceae) popula- selfing genes were transmitted to the offspring. tions with different natural outcrossing rates. Evolution, 44,2031—2046. HOLTSFORD, T. P. AND ELLSTRAND, N. C. 1992. Genetic and Acknowledgements environmental variation in floral traits affecting out- crossing rate in Clarkia tembloriensis (Onagraceae). Iwould like to thank Tammy Cornine, Chris Jung- Evolution, 46, 216—225. bluth, Troy Richey and Kim McCue for their help in JOHNSTON, M. o. 1992. Effects of cross and self-fertiliza- the glasshouse. The comments of Kim McCue, tion on progeny fitness in Lobelia cardinalis and L. Anthony Ippolito and two anonymous reviewers siphilitica. Evolution, 46, 688—702. improved the manuscript. This work was supported JOHNSTON, M. 0. AND SCHOEN, D. j.1994.On the measure- by the University of Missouri Research Board award ment of inbreedingdepression. Evolution,48, 93060. 1735—1741. JOHNSTON, M. 0. AND SCHOEN, D. j.1995.Mutation rates and dominance levels of genes affecting total fitness in References two angiosperm species. Science, 267, 226—229. KARRON, ..D.1989. Breeding systems and the level of AGREN,J. AND SCHEMSKE, D. w. 1993. Outcrossing rate and inbreeding depression in geographically restricted and inbreeding depression in two annual monoecious herbs, widespread species of Astragalus (Fabaceae). Am. J. Begonia hirsuta and B. semiovata. Evolution, 47, Bot., 76, 331—340. 125—135. LANDE, R. AND SCHEMSKE, o. w. 1985. The evolution of BARRETT,S. C. H. AND CHARLESWORTH, D. 1991. Effects of self-fertilization and inbreeding depression in plants. I. a change in the level of inbreeding on the . Genetic models. Evolution, 39, 24—40. Nature, 352, 522—524. LANDE, R., SCHEMSKE, D. W. AND SCHULTZ, S. T. 1994. High BOWMAN, R. N. 1987. Cryptic self-incompatibility and the inbreeding depression, selective interference among breeding system of Ciarkia unguiculata (Onagraceae). loci, and the threshold selfing rate for purging recessive Am. .1. Bot., 74, 471 —476. lethal mutations. Evolution, 48, 965—978. CHARLESWORTH, B., MORGAN, M. T. AND CHARLE5WORTH, LAl-rA, R. AND RITLAND, K. 1994. The relationship between B. 1991. Multilocus models of inbreeding depression inbreeding depression and prior inbreeding among with synergistic selection and partial self-fertilization. populations of four Mimulus taxa. Evolution, 48, Genet. Res., 57, 177—194. 806—817. CHARLESWORTH, D. AND CHARLESWORTFI, B. 1987. Inbreed- LLOYD, D. ci. 1979. Some reproductive factors affecting the ing depression and its evolutionary consequences. Ann. selection of self-fertilization in plants. Am. Nat., 113, Rev. Ecol. Syst., 18, 237—268. 67—79. CHARLESWORTH, D. AND CHARLESWORTH, B. 1990. Inbreed- MAYNARD SMITH, j.1978.The Evolution of Sex. Cambridge ing depression with and its effect on selection for modifiers changing the outcross- University Press, Cambridge, U.K. MCCALL, C., wALLER, D. M. AND MITCHELL-OLD5, T. 1994. ing rate. Evolution, 44, 870—888. Effects of serial inbreeding on fitness components in CHARLESWORTH, D., MORGAN, M. T. AND CHARLESwORTH, Impatiens capensis. Evolution, 48, 818—827. B. 1990. Inbreeding depression, genetic load, and the OUBORG, N. J. AND VAN TRUEREN, R. 1994. The significance evolution of outcrossing rates in a multilocus system of in the process of . IV. with no linkage. Evolution, 44, 1469—1489. Inbreeding load and heterosis in relation to population CROW, i. F. 1948. Alternative hypotheses of vigor. size in the mint Salvia pratensis. Evolution, 48,

The Genetical Society of Great Britain, Heredity, 76, 83—9 1. VARIATION IN INBREEDING DEPRESSION IN CLARK/A 91

996—1008. of self-fertilization and inbreeding depression. III. PRICE, M. V. AND WASER, N. M. 1979. Pollen dispersal and Homozygous lethal mutations at multiple loci. Theor optimal outcrossing in Delphinium nelsoni. Nature, 277, Pop. Biol., 40, 173—210. 294—297. VAN TREUREN, R., BIJLSMA, R., VAN DELDEN, W. AND RATHCKE, B. AND REAL, L. 1993. and inbreeding OUBORG, N. i. 1991. The significance of genetic erosion depression in mountain laurel, Kalmia latifolia (Erica- in the process of extinction. I. Genetic differentiation in ceae).Am. J. Bot., 80, 143—146. Salvia pratensis and Scabiosa columbaria in relation to SAKAI, A. K., KAROLY, K. AND WELLER, S. G. 1989. Inbreed- . Heredity, 66, 181—189. ing depression in Schiedea globosa and S. salicaria (Car- VASEK, F. C. 1964. The evolution of Clarkia unguiculata yophyllaceae), subdioecious and gynodioecious Hawai- derivatives adapted to relatively xeric environments. ian species. Am. J. Bot., 76, 437—444. Evolution, 18, 26—42. SAS INSTITUTE, 1985. SAS User's Guide: Basics and Statis- VASEK, F. C. AND WENG, v. 1988. Breeding systems of tics. SAS Institute, Cary, North Carolina. Clarkia sect. Phaeostoma (Onagraceae): I. Pollen-ovule SCHEMSKE, D. w. 1983. Breeding system and habitat effects ratios. Syst. Bot., 13, 336—350. on fitness components in three rieotropical Costus (Zin- WASER, N. M. 1992. Population structure, optimal out- giberaceae). Evolution, 37, 523—539. breeding, and in angiosperms. In: TOPPINGS, P. 1989. The Significance of Inbreeding Depres- Thornhiil, N. W. (ed.) The Natural Histoty of Inbreeding sion to the Evolution of Self-fertilization in Eichornia and Outbreeding: Theoretical and Empirical Perspectives. paniculata (Spreng.) Solms. (Pontederiaceae). M.S. The- University of Chicago Press, Chicago. sis, University of Toronto. WASER, N. M. AND PRICE, M. v. 1991. Outcrossing distance UYENOYAMA, M. K. AND WALLER, D. M. 1991a. Coevolution effects in Delphinium nelsonii: pollen loads, pollen of self-fertilization and inbreeding depression. I. Muta- tubes, and seed set. Ecology, 72, 171—179. tion-selection balance at one and two loci. Theor Pop. WEIR, B. 5. AND COCKERHAM, C. C. 1973. Mixed self and Biol., 40, 14—46. random mating at two loci. Genet. Res., 21, 247—262. UYENOYAMA, M. K. AND WALLER, D. M. 1991b. Coevolution WILKINSON, L. 1989. SYSTAT. The System for Statistics. of self-fertilization and inbreeding depression. II. Sym- Systat Inc., Evanston, Illinois. metric overdominance in viability. Theor Pop. Biol, 40, WRIGHT, S. 1969. Evolution and Genetics of Populations, 47—77. vol. 2, The Theoty of Gene Frequencies. University of UYENOYAMA, M. K. AND WALLER, D. M. 1991c. Coevolution Chicago Press, Chicago.

The Genetical Society of Great Britain, Heredity, 76, 83—9 1.